AMD workstation market share plunges
Pressure on to get 'Barcelona' out
Requirements Checklist for Choosing a Cloud Backup and Recovery Service Provider
Intel has recaptured all the market share it lost to AMD last year during the two rival chip makers' battle for supremacy in the Windows workstation arena, the latest market stats show.
According to Jon Peddie Research (JPR), AMD's share of the two-socket workstation market slid to just eight per cent during Q1 2007, leaving Intel with a commanding 92 per cent share. Contrast that with the Q1 2006 when AMD's share had almost doubled two quarters to 12.4 per cent.
AMD's market share went on to hit 13.3 per cent, leading many observers to claim the company had at long last got its arch-rival on the run, its Opteron processor finally winning the support Intel's Xeon chip had for so long enjoyed.
Not any more, it seems. While AMD's share fell to 10.9 per cent in Q3 2006, it rallied slightly in Q4, rising to 11.1 per cent, suggesting a short-term decline rather than a systematic plunge. Q1's market drop, however, leaves the company with plenty of lost ground to reclaim.
Include all the single-socket Windows workstations and AMD's share fell from a peak of 3.6 per cent in Q2 2006 to two per cent in Q1 2007.
JPR said some 674,000 dual-socket Windows workstations were sold in Q1, so it's not exactly a colossal market. However, despite over unit sales and shipments rising by, respectively, 15 per cent and 15.2 per cent between Q1 2006 and Q1 2007, shipments of AMD-based workstations were down 24.1 per cent, from 71,040 to 53,920 units.
As JPR researcher Alex Herrera noted, the pressure is now on AMD to get its next-gen quad-core Opeteron chip, 'Barcelona', out the door and prove it's a faster number-cruncher than the latest Xeons.
COMMENTS
Mac Pro
While on the CPU front, the Mac Pro certainly qualifies (I'd say 2 Quad-core Clovertowns is workstation-class in anyone's money), oddly they only offer one workstation-class graphics option. They offer the Mac Pro with a 7300GT (pretty lacklustre, although if you're not doing graphics intensive stuff perfectly fine, and I use one in my work rig currently, albeit that Apple only offer the GDDR2 version rather than the GDDR3-equipped version; an odd choice), a Radeon X1900XT (nice performer, but ATI/AMD's OpenGL performance has never really been up to snuff) or a QuadroFX 4500 (which IS workstation-class, but there're several higher end solutions which they DON'T appear to offer - the 4500X2, 4600, 5500, 5600 or QuadroPlex range). Also, if you're trying to sell a workstation, is posting Doom 3 benchmarks REALLY that relevant - see http://www.apple.com/macpro/graphics.html ?
Mac Pros...
...are hardly the most affordable Macs now, are they?
One wouldn't expect Mac Pros to make up the bulk of Apple's sales, with the Mac Mini and iMacs making up the bulk of the range. I would expect the Mac Pro to be at the narrow end of the sales figures, since buying one involves forking out a rather large sum of money - money that will quite easily buy three or more PCs that are far more suitable for the average computer user (and are, incidentally, about as expandable as a Mac Pro). Therefore, less people are likely to buy a Mac Pro.
AMD are hardly likely to be worried about the Mac Pro's sales. What's likely to be more worrying to them is this whole shift to the new AM2 socket. That's immediately lost them a lot of loyal customers, as it alienated existing users with no furture upgrade path and gave Intel a huge advantage. AMD always made its case on being enthusiast-friendly, and they scored a *huge* own-goal when they switched everything to AM2. AMD's customers are typically not very affluent, and usually chose AMD because they were the cheapest - so angering these people by forcing them to change platforms was not a wise move. They won't switch back in a hurry, even if Barcelona is the best thing since sliced bread - since they've just emptied their wallets buying Intel kit. By the time they next upgrade, Nehalem will be in every computer shop.
This is just the beginning. Unfortunately, even if Barcelona really kicks ass, I don't think it will do much for AMD. Intel has woken up to the importance of power management with the Core 2 series - and it's now being taken seriously by corporates again. AMD is effectively starting from square one again, against a bigger rival with much more resources to pour into chipset development (one of the key reasons I have always bought into the Intel platform - CPU is always second in importance for me; reliability comes first). Yorkfield will also do what AMD's been promising, and it will launch in time to really pour a lot of rain on AMD's parade ... and the forthcoming Nehalem will dispense with FSB altogether and move the data-starved Intel platform onto new levels of performance.
AMD had better pull its socks up. Barcelona will, at best, keep them in the game for now. They need to produce something positively spectacular to compete against Nehalem. That's what's really going to kill them.
Oliver.
Unfortunate, but no surprise...
AMD has always been behind Intel on process transitions. Every time Intel has made an aggressive shift to a new process node AMD has trailed by half a year or more. And everytime AMD has been under extreme pressure just about the time it manages to make the transition.
Intel's current aggressive push with Core1 and then Core2 at 65nm has put AMD in the place it's been repeatedly before. Folks might remember the K6 and the K7 -- in both cases AMD was having so much trouble with process that it had to turn to IBM for fab to get through.
Barcelona doesn't look to be in that kind of trouble ... yet anyway.
IT infrastructure monitoring strategies
What you need to know about cloud backup
Enabling efficient data center monitoring
Agentless Backup is Not a Myth
Top 10 SIEM Implementer’s Checklist
